Cochlear Hearing Loss: Physiological, Psychological and Technical Issues

COCHLEARHEARING LOSSPhysiological, Psychological andTechnical Issues

Second Edition

BRIAN C. J. MOOREDepartment of Experimental Psychology,University of Cambridge,Downing Street, Cambridge CB2 3EB,England

COCHLEAR HEARING LOSS

COCHLEARHEARING LOSSPhysiological, Psychological andTechnical Issues

Second Edition

BRIAN C. J. MOOREDepartment of Experimental Psychology,University of Cambridge,Downing Street, Cambridge CB2 3EB,England

Copyright 2007 John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester,West Sussex PO19 8SQ, England

Telephone (+44) 1243 779777

Email (for orders and customer service enquiries): [email protected] our Home Page on www.wiley.com

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in anyform or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms ofthe Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing AgencyLtd, 90 Tottenham Court Road, London W1T 4LP, UK, without the permission in writing of the Publisher. Requests tothe Publisher should be addressed to the Permissions Department, John Wiley & Sons Ltd, The Atrium, Southern Gate,Chichester, West Sussex PO19 8SQ, England, or emailed to [email protected], or faxed to (+44) 1243 770620.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand names andproduct names used in this book are trade names, service marks, trademarks or registered trademarks of their respectiveowners. The Publisher is not associated with any product or vendor mentioned in this book.

This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. Itis sold on the understanding that the Publisher is not engaged in rendering professional services. If professional adviceor other expert assistance is required, the services of a competent professional should be sought.

Other Wiley Editorial Offices

John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA

Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA

Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany

John Wiley & Sons Australia Ltd, 42 McDougall Street, Milton, Queensland 4064, Australia

John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809

John Wily & Sons Canada Ltd, 6045 Freemont Blvd, Mississauga, ONT, L5R 4J3, Canada

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be availablein electronic books.

Anniversary Logo Design: Richard J. Pacifico

Library of Congress Cataloging-in-Publication DataMoore, Brian C. J.Cochlear hearing loss : physiological, psychological and technical issues / Brian C.J. Moore. - - 2nd ed.

p. ; cm.Includes bibliographical references and index.ISBN 978-0-470-51633-1 (alk. paper)1. Hearing disorders. 2. Cochlea- -Pathophysiology. 3. Hearing- -Physiological aspects. I. Title.[DNLM: 1. Hearing Disorders. 2. Cochlea- -physiopathology. 3. Hearing- -physiology. WV 270 M821c 2007]RF291.M658 2007617.882dc22

2007015880British Library Cataloguing in Publication DataA catalogue record for this book is available from the British Library

ISBN: 978-0-470-51633-1

Typeset by Thomson Press (India) Limited, IndiaPrinted and bound in Great Britain by TJ International, Padstow

This book is printed on acid-free paper responsibly manufactured from sustainable forestryin which at least two trees are planted for each one used for paper production.

Contents

Preface xi

1 Physiological Aspects of Cochlear Hearing Loss 1

I. INTRODUCTION 1II. LINEAR AND NONLINEAR SYSTEMS 1

III. STRUCTURE AND FUNCTION OF THE OUTER ANDMIDDLE EAR 6

IV. STRUCTURE AND FUNCTION OF THE NORMAL COCHLEA 91. The cochlea, the basilar membrane and the organ of Corti 92. Tuning on the basilar membrane 123. The nonlinearity of input-output functions on the basilar membrane 164. Two-tone suppression 185. Combination tone generation 186. Responses of the basilar membrane to complex sounds 197. Otoacoustic emissions 20

V. NEURAL RESPONSES IN THE NORMAL AUDITORY NERVE 211. Spontaneous firing rates and thresholds 222. Tuning curves and iso-rate contours 223. Rate-versus-level functions 234. Two-tone suppression 255. Phase locking 26

VI. TYPES OF HEARING LOSS 28VII. PHYSIOLOGY OF THE DAMAGED COCHLEA 29

1. Basilar membrane responses 292. Neural responses 313. Structurefunction correlation 324. Otoacoustic emissions 355. Phase locking 35

VIII. CONCLUSIONS 36

2 Absolute Thresholds 39

I. INTRODUCTION 39II. MEASURES OF ABSOLUTE THRESHOLD 39

1. Minimum audible pressure (MAP) 392. Minimum audible field (MAF) 393. Comparison of MAP and MAF 404. The audiogram 41

III. DESCRIPTIONS OF THE SEVERITY OF HEARING LOSS 42IV. CAUSES OF HEARING LOSS DUE TO COCHLEAR DAMAGE 43V. PERCEPTUAL CONSEQUENCES OF ELEVATED ABSOLUTE

THRESHOLDS 44

vi CONTENTS

3 Masking, Frequency Selectivity and Basilar Membrane Nonlinearity 45

I. INTRODUCTION 45II. THE MEASUREMENT OF FREQUENCY SELECTIVITY USING

MASKING 461. Introduction 462. The power-spectrum model 463. Estimating the shape of a filter 47

III. ESTIMATING FREQUENCY SELECTIVITY FROM MASKINGEXPERIMENTS 481. Psychophysical tuning curves 482. The notched-noise method 51

IV. CHARACTERISTICS OF THE AUDITORY FILTER IN NORMALHEARING 541. Variation with centre frequency 542. Variation with level 563. Summary 59

V. MASKING PATTERNS AND EXCITATION PATTERNS 591. Masking patterns 592. Relationship of the auditory filter to the excitation pattern 613. Changes in excitation patterns with level 624. Possible effects of suppression 63

VI. NON-SIMULTANEOUS MASKING 641. Basic properties of non-simultaneous masking 642. Evidence for suppression from non-simultaneous masking 673. The enhancement of frequency selectivity revealed in non-simultaneous

masking 694. Relation between the growth of forward masking and the basilar

membrane input-output function 70VII. THE AUDIBILITY OF PARTIALS IN COMPLEX TONES 73

VIII. EFFECTS OF COCHLEAR DAMAGE ON FREQUENCYSELECTIVITY IN SIMULTANEOUS MASKING 751. Complicating factors 752. Psychophysical tuning curves 763. Auditory filter shapes measured with notched noise 79

IX. THE USE OF MASKING TO DIAGNOSE DEAD REGIONS 831. The threshold-equalizing noise (TEN) test 832. The TEN(HL) test 853. Prevalence of dead regions assessed using the TEN(HL) test 86

X. EFFECTS OF COCHLEAR DAMAGE ON FORWARD MASKINGAND SUPPRESSION 86

XI. EFFECTS OF COCHLEAR HEARING LOSS ON BMINPUT-OUTPUT FUNCTIONS 88

XII. PERCEPTUAL CONSEQUENCES OF REDUCEDFREQUENCY SELECTIVITY, DEAD REGIONS,LOSS OF SUPPRESSION AND STEEPER BMINPUT-OUTPUT FUNCTIONS 901. Susceptibility to masking 902. Timbre perception 903. Perceptual consequences of dead regions 91

CONTENTS vii

4 Loudness Perception and Intensity Resolution 93

I. INTRODUCTION 93II. LOUDNESS PERCEPTION FOR NORMALLY

HEARING PEOPLE 931. Equal-loudness contours and loudness level 932. The scaling of loudness 943. The detection of intensity changes 96

III. EFFECTS OF COCHLEAR HEARING LOSS ONLOUDNESS PERCEPTION 97

IV. A MODEL OF NORMAL LOUDNESS PERCEPTION 101V. A MODEL OF LOUDNESS PERCEPTION APPLIED TO COCHLEAR

HEARING LOSS 1041. Introduction 1042. Elevation of absolute threshold 1053. Reduced compressive nonlinearity 1054. Reduced inner hair cell/neural function 1065. Reduced frequency selectivity 1076. Complete loss of functioning IHCs or neurones (dead regions) 1087. Using the model to account for loudness recruitment 109

VI. EFFECTS OF BANDWIDTH ON LOUDNESS 1101. Normal hearing 1102. Impaired hearing 111

VII. EFFECTS OF COCHLEAR HEARING LOSS ONINTENSITY RESOLUTION 113

VIII. PERCEPTUAL CONSEQUENCES OF ALTEREDLOUDNESS PERCEPTION 1141. Consequences of loudness recruitment and reduced

dynamic range 1142. Perceptual consequences of reduced loudness summation 1143. Perceptual consequences of altered intensity discrimination 115

5 Temporal Resolution and Temporal Integration 117

I. INTRODUCTION 117II. MODELLING WITHIN-CHANNEL TEMPORAL RESOLUTION IN

NORMAL HEARING 1181. Bandpass filtering 1182. The nonlinearity 1193. The sliding temporal integrator 1204. The decision device 1225. Characterizing the nonlinear device and the sliding

temporal integrator 122III. TEMPORAL RESOLUTION IN NORMAL HEARING 124

1. The effect of centre frequency on gap detection 1242. Temporal modulation transfer functions 1253. The rate of recovery from forward masking 126

IV. TEMPORAL RESOLUTION IN PEOPLE WITHCOCHLEAR DAMAGE 1281. The influence of sound level on gap detection and the rate of decay of

forward masking 128

viii CONTENTS

2. The influence of audible bandwidth on temporal modulation transferfunctions and gap detection 130

3. The influence of changes in the compressive nonlinearity 131V. TEMPORAL INTEGRATION AT THRESHOLD 135

1. Temporal integration in normally hearing people 1352. Temporal integration in people with cochlear hearing loss 1363. Explanations for reduced temporal integration in people with cochlear

hearing loss 137VI. TEMPORAL INTEGRATION AT SUPRATHRESHOLD LEVELS 138

VII. PERCEPTUAL CONSEQUENCES OF ABNORMAL TEMPORALPROCESSING IN PEOPLE WITH COCHLEAR HEARING LOSS 1401. Consequences of abnormal temporal resolution 1402. Consequences of reduced temporal integration 141

6 Pitch Perception and Frequency Discrimination 143

I. INTRODUCTION 143II. THEORIES OF PITCH PERCEPTION 144

III. THE PERCEPTION OF THE PITCH OF PURE TONES BY NORMALLYHEARING PEOPLE 1441. The frequency discrimination of pure tones 1442. The perception of musical intervals 1483. The effect of level on pitch 149

IV. FREQUENCY DISCRIMINATION OF PURE TONES BY PEOPLEWITH COCHLEAR HEARING LOSS 1501. Difference limens for frequency (DLFs) 1502. Frequency modulation detection limens (FMDLs) 152

V. THE PERCEPTION OF PURE-TONE PITCH FOR FREQUENCIESFALLING IN A DEAD REGION 155

VI. PITCH ANOMALIES IN THE PERCEPTION OF PURE TONES 157VII. THE PITCH PERCEPTION OF COMPLEX TONES BY NORMALLY

HEARING PEOPLE 1591. The phenomenon of the missing fundamental 1592. Discrimination of the repetition rate of complex tones 159

VIII. THEORIES OF PITCH PERCEPTION FOR COMPLEX TONES 1601. The representation of a complex tone in the peripheral

auditory system 1602. Spectro-temporal pitch theories 1623. The relative importance of envelope and temporal fine structure 164

IX. PITCH PERCEPTION OF COMPLEX TONES BY PEOPLE WITHCOCHLEAR HEARING LOSS 1671. Theoretical considerations 1672. Experimental studies 169

X. PERCEPTUAL CONSEQUENCES OF ALTERED FREQUENCYDISCRIMINATION AND PITCH PERCEPTION 1701. Effects on speech perception 1702. Effects on music perception 172

7 Spatial Hearing and Advantages of Binaural Hearing 173

I. INTRODUCTION 173II. THE LOCALIZATION OF SINUSOIDS 174

1. Cues for localization 174

CONTENTS ix

2. Performance of normally hearing people in localizationand lateralization 177

3. Performance of hearing-impaired people in localization andlateralization 178

III. THE LOCALIZATION OF COMPLEX SOUNDS 1791. The role of transients and across-frequency comparisons 1792. Performance of normally hearing people 1793. Performance of people with cochlear hearing loss 1804. Reasons for large interaural time difference and interaural level

difference thresholds in people with cochlear hearing loss 183IV. THE CONE OF CONFUSION, HEAD MOVEMENTS AND

PINNA CUES 1841. The cone of confusion 1842. The role of head movements 1853. Information provided by the pinnae 1854. Localization using pinna cues by normally hearing and hearing-impaired

people 186V. GENERAL CONCLUSIONS ON SOUND LOCALIZATION 186

VI. THE PRECEDENCE EFFECT 1871. The precedence effect for normal hearing 1872. The precedence effect for impaired hearing 188

VII. BINAURAL MASKING LEVEL DIFFERENCES (MLDs) 1891. MLDs for normally hearing people 1892. Mechanisms underlying MLDs 1923. MLDs for people with cochlear hearing loss 1924. Possible reasons for smaller MLDs in people with

cochlear damage 193VIII. HEAD-SHADOW EFFECTS 194

1. Benefits of head shadow for normally hearing people 1942. Benefits of head shadow for hearing-impaired people 195

IX. RELEASE FROM INFORMATIONAL MASKING 196X. DIOTIC ADVANTAGES 198

XI. PERCEPTUAL CONSEQUENCES OF ABNORMAL BINAURAL ANDSPATIAL HEARING IN PEOPLE WITH COCHLEAR DAMAGE 199

8 Speech Perception 201

I. INTRODUCTION 201II. THE MAGNITUDE OF THE NOISE PROBLEM 201

III. THE ROLE OF AUDIBILITY 2031. The Articulation Index (AI) and Speech Intelligibility Index (SII) 2032. Use of the AI or SII to predict speech intelligibility for the hearing

impaired 2043. The intelligibility of speech in noise at high overall levels 2054. Comparison of detection and recognition for speech in noise 2065. The intelligibility of speech in quiet at high overall levels 2076. Simulation of hearing loss by selective filtering (frequency-dependent

attenuation) 2077. Simulation of hearing loss by masking 2088. Conclusions on the role of audibility 209

IV. INFLUENCE OF DEAD REGIONS ON SPEECH PERCEPTION 209V. CORRELATION BETWEEN PSYCHOACOUSTIC ABILITIES AND

SPEECH PERCEPTION 212

x CONTENTS

VI. ASSESSING THE EFFECTS OF FREQUENCY SELECTIVITY ONVOWEL AND CONSONANT PERCEPTION 2141. Consonant perception 2142. Vowel perception 215

VII. INFLUENCE OF LOSS OF SENSITIVITY TO TEMPORAL FINESTRUCTURE 219

VIII. THE USE OF SIMULATIONS TO ASSESS THE IMPORTANCE OFPSYCHOACOUSTIC FACTORS IN SPEECH PERCEPTION 2211. Simulations of loudness recruitment combined with

threshold elevation 2222. Simulations of reduced frequency selectivity 2263. Simulation of the combined effects of threshold elevation, recruitment

and reduced frequency selectivity 2294. Simulation of reduced temporal resolution 230

IX. CONCLUSIONS 232

9 Hearing Aids 233

I. INTRODUCTION 233II. LINEAR AMPLIFICATION 233

1. The difficulty of restoring audibility using linear aids 2332. Prescriptive fitting rules for linear hearing aids 234

III. COMPRESSION AMPLIFICATION 2361. Basic characteristics of automatic gain control systems 2362. Varieties of automatic gain control systems 2413. Rationales for the use of multi-band compression

(and noise reduction) 2414. Research on the effectiveness of multi-band syllabic compression 2425. Methods for initial fitting of hearing aids with

multi-band compression 2446. Methods for fine tuning hearing aids with multi-band compression 2527. Slow-acting automatic gain control systems 2538. Comparisons of slow-acting and fast-acting systems 2559. General conclusions about compression 257

IV. SOME GENERAL PROBLEMS WITH HEARING AIDS 2571. Inadequate gain at high frequencies 2572. Acoustic feedback 2583. Peakiness of frequency response 2594. The occlusion effect 2605. Time delays 261

V. METHODS FOR IMPROVING THE SPEECH-TO-NOISE RATIO 2621. Multi-channel noise reduction 2622. Directional microphones 2623. Binaural processing algorithms 263

VI. TRANSPOSITION AIDS FOR SEVERE AND PROFOUND HEARINGLOSS 264

VII. COCHLEAR IMPLANTS 266VIII. CONCLUDING REMARKS 267

Glossary 269

References 287

Index 327

Preface

The is a second edition of Cochlear Hearing Loss, which was first published in 1998.The book is intended to impart an understanding of the changes in perception that takeplace when a person has a cochlear hearing loss. I have tried to interrelate physiologicaldata and perceptual data. The book aims to present both data and concepts in anintegrated way. The main goal is to convey an understanding of the perceptual changesassociated with cochlear hearing loss, of the difficulties faced by the hearing-impairedperson and of the possibilities and limitations of current hearing aids. The readershould come away with an impression not only of what happens but also of why ithappens. I have attempted to write at a level that is accessible and comprehensible tostudents in audiology and in speech and hearing science. I hope, therefore, that thebook will be considered suitable as a textbook for such students.

The book assumes some prior knowledge about sound. For example, it assumesfamiliarity with the decibel and the concept of the spectrum of a sound. However, anextensive glossary is provided to give brief explanations of technical terms used in thebook. Also the concept of linear and nonlinear systems is explained in some detail,as this concept is crucial for understanding this book, but it is often poorly explainedin textbooks on hearing.

Appreciation of the perceptual effects of cochlear hearing loss can be helped bylistening to simulations of these effects. A compact disc containing simulations ofdifferent aspects of cochlear damage (Moore, 1997) can be obtained by writing tothe author at: Department of Experimental Psychology, University of Cambridge,Downing Street, Cambridge CB2 3EB, England. The cost of the disc is 12 poundssterling or 20 US dollars. Payment (a cheque payable to B.C.J. Moore) should beenclosed with the order. The disc also illustrates the effectiveness of various types ofamplification and compression in compensating for the effects of cochlear hearingloss.

This second edition has been extensively updated and revised. In particular, newmaterial has been added on: dead regions in the cochlea, pitch perception, loudnessperception, the precedence effect, informational masking, speech perception and hear-ing aids.

I would like to thank the colleagues who commented on preliminary drafts of partsof the first or second editions of this book. They include Jose Alcantara, Thomas Baer,Deniz Baskent, Peter Blamey, Brent Edwards, Brian Glasberg, Hedwig Gockel, ErvinHafter, Robert Peters, Marina Rose, Aleksander Sek, Michael Stone, Deborah Vickers,Joyce Vliegen and Magdalena Wojtczak. I would also like to thank Ian Cannell,Brian Glasberg, Aleksander Sek and Michael Stone for considerable assistance inproducing figures. The second edition was written while I was on sabbatical leave

xii PREFACE

at the Department of Psychology, University of California at Berkeley. I thank ErvinHafter for inviting me to come and for the use of his office, and Lee Hafter for manysplendid dinners.

Brian C.J. MooreMarch, 2007

1 Physiological Aspects of CochlearHearing Loss

I INTRODUCTION

Hearing loss caused by damage to the cochlea is probably the most common form ofhearing loss in the developed countries. Its most obvious symptom, and the one thatis almost always assessed in the clinic, is an elevation of the threshold for detectingsounds. However, it is also accompanied by a variety of other changes in the way thatsound is perceived. Even if sounds are amplified (e.g. by a hearing aid) so that theyare well above the threshold for detection, the perception of those sounds is usuallyabnormal; the person with cochlear hearing loss often reports that the sounds areunclear and distorted, and that it is hard to hear comfortably over a wide range ofsound levels. A common complaint is difficulty in understanding speech, especiallywhen background sounds or reverberation are present. One of the main aims of thisbook is to explain why these problems occur and why current hearing aids are oflimited benefit in compensating for the problems.

The book assumes that the reader has a basic knowledge of physics and acoustics,for example an understanding of what is meant by terms such as sinusoid, spectrum,frequency component and the decibel. The reader who is not familiar with these termsshould consult a textbook, such as An Introduction to the Psychology of Hearing(Moore, 2003) or Signals and Systems for Speech and Hearing (Rosen and Howell,1991). For the reader who knows these things, but needs a reminder, many of thekey terms are defined briefly in the Glossary. Most of the terms that appear in italicsin the text are defined in the Glossary. One concept that may not be familiar is thatof a linear system. This topic is of importance, since the normal peripheral auditorysystem shows significant nonlinearities, whereas the system becomes more linearwhen cochlear damage occurs. Hence, this chapter starts with a description of theproperties of linear and nonlinear systems. It then goes on to consider the physiologyand the function of the normal and damaged cochlea.

II LINEAR AND NONLINEAR SYSTEMS

The auditory system is often thought of as a series of stages, the output of a given stageforming the input to the next. Each stage can be considered as a device or system,with an input and an output. For a system to be linear, certain relationships betweenthe input and output must hold true. The following two conditions must be satisfied:

2 COCHLEAR HEARING LOSS

1. If the input to the system is changed in magnitude by a factor k, then the outputshould also change in magnitude by a factor k, but be otherwise unaltered. Thiscondition is called homogeneity. For example, if the input is doubled, then theoutput is doubled, but without any change in the form of the output. Thus, a plotof the output as a function of the input would be a straight line passing throughthe origin (zero input gives zero output) hence the term linear system. Such aplot is called an input-output function. An example of such a function is givenin panel (a) of Figure 1.1.

2. The output of the system in response to a number of independent inputs pre-sented simultaneously should be equal to the sum of the outputs that would havebeen obtained if each input were presented alone. For example, if the response

Figure 1.1. The left column shows input-output functions for a linear system (a), a nonlinearsystem with hard peak clipping (d) and a nonlinear system with more progressive saturation(g). The middle column shows outputs from these systems in response to a sinusoidal input.The third column shows spectra of the outputs.

PHYSIOLOGICAL ASPECTS OF COCHLEAR HEARING LOSS 3

to input A is X and the response to input B is Y, then the response to A and Btogether is simply X + Y. This condition is known as superposition.

When describing a system as linear, it is usually assumed that the system is time-invariant. This means that the input-output function does not change over time. Forexample, if the input is I and the output is O, the relationship between the input andthe output would be:

O = cI, (1.1)

where c is a constant that does not vary with time.When a sinusoid is used as an input to a linear system, the output is a sinusoid

of the same frequency. This is illustrated in panel (b) of Figure 1.1. The amplitudeand phase of the output may, however, be different from those of the input. Assumethat the input is a single sinusoid whose waveform as a function of time, I(t), can bedescribed by:

I(t) = Asin(2ft), (1.2)

where A is the peak amplitude of the input and f is the frequency in Hz (cycles persecond). The output as a function of time, O(t), could then be represented by:

O(t) = G Asin(2ft + ), (1.3)

where G is a constant representing the amplitude ratio between the input and output,and is a constant representing the phase shift between the input and output. G issometimes referred to as the gain of the system.

Since the output of a linear system in response to a sinusoidal input is itselfsinusoidal, the spectrum of the output, by definition, consists of a single frequencycomponent. This is illustrated in panel (c) of Figure 1.1. More generally, the output ofa linear system never contains frequency components that were not present in the inputsignal. The response of a linear system may, however, vary with the frequency of theinput sinusoid. This is equivalent to saying that the constants G and in Equation 1.3can vary with the input frequency.

In practice, many devices or systems are linear as long as the input is not too large.Excessive inputs may cause the system to become nonlinear; more details are givenlater on. Such a system is usually called linear, even though it can become nonlinearunder extreme conditions. As an example, consider a loudspeaker. The input is a volt-age and the output is a movement of the cone of the loudspeaker, which can produceaudible sound waves. For the types of inputs that are typically used for a loudspeaker,the response is approximately linear; the conditions of homogeneity and superpositionare obeyed. If the input voltage varies in a sinusoidal manner, the movement of thecone is almost sinusoidal. However, if the frequency of the input is changed, holdingthe magnitude of the input constant, the magnitude of the movement of the cone may


vary. In this case, we would say that the loudspeaker does not have a flat frequencyresponse; G may vary with frequency. Similarly, may vary with frequency. Otherexamples of systems that usually operate in a nearly linear way are microphones,amplifiers and the output transducers used in hearing aids (often called receivers).

When waveforms other than sinusoids are applied as the input to a linear system,the output waveform will often differ from that of the input. For example, if theinput to a linear system is a square wave, the output is not necessarily a square wave.This is one reason for the popularity of sinusoids in auditory research; sinusoids arethe only waveforms which are always preserved by a linear system. If a system islinear, then it is relatively easy to predict its output for any arbitrary complex input.As a first step, the output is measured as a function of frequency for a sinusoidalinput. Essentially, the values of G and are determined as a function of the inputfrequency. To predict the output for a complex input, a Fourier analysis of theinput is performed. This gives a description of the input in terms of the amplitudesand phases of its sinusoidal components. The output for each of the sinusoidalcomponents comprising the input can then be calculated. Finally, using the principleof superposition, the output in response to the whole complex can be calculated asthe sum of the outputs in response to its individual sinusoidal components. This isa powerful method, and it gives another reason for using sinusoids as stimuli.

As mentioned earlier, many linear systems become nonlinear if the input is madelarge enough. An example is shown in panel (d) of Figure 1.1. The input-outputfunction is linear over a large part of its range, but it flattens out for large positiveor negative values of the input. This is sometimes called saturation or peak clippingand can occur in systems such as transistor amplifiers and condenser microphones. Inthe example shown, the clipping is symmetrical, in that it occurs at the same absolutevalue for positive and negative values of the input. When a sinusoid is used as inputto such a system and the peak amplitude of the sinusoid, A, is sufficiently large,the output is no longer sinusoidal. This is illustrated in panel (e) of Figure 1.1. Theoutput is periodic, with the same period as the input sinusoid, but the waveform isdistorted. In this case, the output contains frequency components that are not presentin the input. This is illustrated in panel (f) of Figure 1.1. In general, the output of anonlinear system in response to a single sinusoid at the input contains one or morecomponents (sinusoids) with frequencies that are integer multiples of the frequency ofthe input. These components are referred to as harmonics, and the nonlinear system issaid to introduce harmonic distortion. For example, if the input was a sinusoid with afrequency of 500 Hz, the output might still contain a component with this frequency,but components with other frequencies might be present too, for example 1000, 1500,2000 . . . Hz.

Another example of a nonlinear input-output function is shown in panel (g) ofFigure 1.1. In this case, the function does not show hard clipping, but the slopebecomes more shallow when the absolute value of the input or output exceeds acertain value. This type of input-output function can occur in valve (tube) amplifiers,moving coil microphones and loudspeakers, and it can also occur in the auditorysystem. The output waveform, shown in panel (h) of Figure 1.1, is less distorted than


when hard clipping occurs, and the output spectrum, shown in panel (i), reveals lessharmonic distortion.

If the input to a nonlinear system consists of two sinusoids, then the output may con-tain components with frequencies corresponding to the sum and difference of the twoinput frequencies, and their harmonics, as well as the original sinusoidal componentsthat were present in the input. These extra components are said to result from inter-modulation distortion. For example, if the input contains two sinusoids with frequen-cies f1 and f2, the output may contain components with frequencies f1 f2, f1 + f2,2f1 f2, 2f2 f1 and so on. These components are referred to as intermodulationdistortion products, and, in the case of the auditory system, they are also calledcombination tones.

When a system is nonlinear, the response to complex inputs cannot generally bepredicted from the responses to the sinusoidal components comprising the inputs.Thus, the characteristics of the system must be investigated using both sinusoidal andcomplex inputs.

Often, the input and output magnitudes of a system are plotted on logarithmic axes(the decibel scale is an example). In that case, the input and output magnitudes arenot specified as instantaneous values (e.g. as the instantaneous voltage in the caseof an electrical signal). Generally, instantaneous magnitudes can have both positiveand negative values, but it is not possible to take the logarithm of a negative number.Instead, the input and output are averaged over a certain time, and the magnitudeis expressed as a quantity which cannot have a negative value. Typically, both theinput and output are specified in terms of their power (related to the mean-squarevalue) or their root-mean-square value. Sometimes, the peak amplitude may be used.For a linear system, the condition of homogeneity still applies to such measures. Forexample, if the input power is doubled, the output power is also doubled. When plottedon log-log axes, the input-output function of a linear system is a straight line witha slope of unity. To see why this is the case, we take the logarithm of both sides ofEquation 1.1 (O = cI). This gives:

log(O) = log(cI) = log(c) + log(I) (1.4)

The value of log(c) is itself a constant. Therefore, since log(O) is simply equal tolog(I) plus a constant, the slope of the line relating log(O) to log(I) must be unity. Ina nonlinear system, the slope of the input-output function on logarithmic axes differsfrom unity. Say, for example, that the output is proportional to the square of the input:

O = cI2 (1.5)

Taking logarithms of both sides gives:

log(O) = log(cI2) = log(c) + 2log(I) (1.6)


In this case, the slope of the input-output function on logarithmic axes is two. When theslope of the input-output function is greater than one, the nonlinearity is referred to asexpansive. If the output were proportional to the square-root of the input (O = cI0.5),the slope of the function would be 0.5. When the slope of the input-output function isless than one, the nonlinearity is referred to as compressive. Examples of input-outputfunctions plotted on log-log axes will be presented later in this chapter, in connectionwith the response of the basilar membrane (BM) within the cochlea.

III STRUCTURE AND FUNCTION OF THE OUTERAND MIDDLE EAR

Figure 1.2 shows the structure of the peripheral part of the human auditory system. Itis composed of three parts, the outer, middle and inner ear. The outer ear includes thepinna and the auditory canal, or meatus. The pinna and meatus together create a broadresonance which enhances sound levels at the eardrum, relative to those obtained inthe absence of the listeners head, over the frequency range from about 1.5 to 5 kHz.

Figure 1.2. Illustration of the structure of the peripheral auditory system showing the outer,middle and inner ear. Redrawn from Lindsay and Norman (1972).


Figure 1.3. The difference between the sound level measured at the eardrum (for a soundcoming from the frontal direction) and the sound level in the free field at the point correspondingto the centre of the listeners head. Data from Shaw (1974).

This is illustrated in Figure 1.3. The maximum boost is typically about 1215 dB inthe region around 2.5 kHz. When a hearing aid is fitted so as to block the meatus, thisbroad resonance is lost for a behind-the-ear aid, reduced in magnitude for a conchaaid, and shifted to higher frequencies for a completely-in-the-canal aid. This has to betaken into account when fitting hearing aids and especially when adjusting the gainas a function of frequency.

At medium and high frequencies, the sound reaching the eardrum is significantlymodified by the pinna, head and upper torso. Specifically, when the sound containsa broad range of frequencies, the pinna introduces a complex pattern of peaks andnotches into the spectrum. This pattern varies systematically with the direction of thesound source relative to the head, and the spectral patterns thus provide importantinformation about the location of sound sources (see Chapter 7 for more details).

Sound travels down the meatus and causes the eardrum, or tympanic membrane, tovibrate. The eardrum forms the outer boundary of the middle ear. These vibrations aretransmitted through the middle ear by three small bones, the ossicles, to a membrane-covered opening in the bony wall of the spiral-shaped structure of the inner ear thecochlea. This opening is called the oval window and it forms the outer boundary ofthe middle ear. The three bones are called the malleus, incus and stapes (popularly


known as the hammer, anvil and stirrup), the stapes being the lightest and smallest ofthese and the one which actually makes contact with the oval window.

The major function of the middle ear is to ensure the efficient transfer of soundenergy from the air to the fluids in the cochlea. If the sound were to impinge di-rectly onto the oval window, most of it would simply be reflected back, rather thanentering the cochlea. This happens because the resistance of the oval window to move-ment is very different from that of air. This is described as a difference in acousticalimpedance. The middle ear acts as an impedance-matching device or transformerthat improves sound transmission and reduces the amount of reflected sound. Thisis accomplished mainly by the 27 : 1 ratio of effective areas of the eardrum and theoval window, and to a small extent by the lever action of the ossicles. Transmission ofsound energy through the middle ear is most efficient at middle frequencies (5005000Hz), which are the ones most important for speech perception; see Chapter 8. Thisis illustrated in Figure 1.4, adapted from Glasberg and Moore (2006), which showsan estimate of the relative effectiveness of transmission through the middle ear as afunction of frequency.

The ossicles have minute muscles attached to them which contract when we areexposed to intense sounds. This contraction, known as themiddle ear reflex or acousticreflex, is probably mediated by neural centres in the brain stem (Liberman and Guinan,1998). The reflex can be triggered by sound of any frequency, but it reduces thetransmission of sound through the middle ear only at low frequencies (below about1000 Hz). It may help to prevent damage to the delicate structures inside the cochlea.However, the activation of the reflex is too slow to provide any protection againstimpulsive sounds, such as gunshots or hammer blows.

Figure 1.4. The transfer function of the middle ear, plotted as relative response versus fre-quency. The response was arbitrarily labelled as 2.6 dB at 1 kHz. The estimate comes fromGlasberg and Moore (2006).


For moderate sound levels (below about 90 dB SPL), the outer and middle earbehave essentially as linear systems; they do not introduce significant harmonic orintermodulation distortion. However, at high sound levels both the tympanic mem-brane and the ossicles may vibrate in a nonlinear manner, and the acoustic reflexalso introduces nonlinearity. These nonlinearities may result in audible harmonic andintermodulation distortion.

IV STRUCTURE AND FUNCTION OF THE NORMALCOCHLEA

IV.1 THE COCHLEA, THE BASILAR MEMBRANE AND THE ORGANOF CORTI

The inner ear is also known as the cochlea. It is shaped like the spiral shell of a snail.However, the spiral shape does not appear to have any functional significance, and thecochlea is often described as if the spiral had been unwound. The cochlea is filledwith almost incompressible fluids, and it has bony rigid walls. It is divided alongits length by two membranes, Reissners membrane and the BM (see Figure 1.5).

Figure 1.5. Cross-section of the cochlea, showing the BM, Reissners membrane and the organof Corti. Redrawn from Davis (1962).


The start of the spiral, where the oval window is situated, is known as the base; theother end, the inner tip, is known as the apex. It is also common to talk about thebasal end and the apical end. At the apex there is a small opening (the helicotrema)between the BM and the walls of the cochlea which connects the two outer chambersof the cochlea, the scala vestibuli and the scala tympani. Inward movement of theoval window results in a corresponding outward movement in a membrane covering asecond opening in the cochlea the round window. Such movements result in pressuredifferences between one side of the BM and the other (i.e. the pressure is applied ina direction perpendicular to the BM), and this results in movement of the BM (seebelow for details). The helicotrema eliminates any pressure differences between thescala vestibuli and the scala tympani at very low frequencies. This prevents the BMfrom moving significantly in response to movements of the oval window caused byjaw movements or by slow changes in air pressure (such as occur when changingaltitude). The helicotrema also reduces movement of the BM in response to low-frequency sounds.

On the side of the cochlea closest to the outer wall (the right-hand side in Figure1.5), there is a structure called the stria vascularis. This plays a strong role in themetabolism of the cochlea and in creating the voltages (electrical potentials) that areessential for the normal operation of the cochlea. The stria vascularis is sometimescolloquially described as the battery of the cochlea.

A third membrane, called the tectorial membrane, lies above the BM, and alsoruns along the length of the cochlea. Between the BM and the tectorial membraneare hair cells, which form part of a structure called the organ of Corti (see Fig-ures 1.5 and 1.6). They are called hair cells because they appear to have tufts of

Figure 1.6. Cross section of the organ of Corti as it appears in the basal turn of the cochlea.


hairs, called stereocilia, at their apexes. The hair cells are divided into two groupsby an arch known as the tunnel of Corti. Those on the side of the arch closest tothe outside of the spiral shape are known as outer hair cells (OHCs), and they arearranged in three rows in cats and up to five rows in humans, although the rowsare often somewhat irregular in humans (Wright et al., 1987). The hair cells onthe other side of the arch form a single row, and are known as inner hair cells(IHCs). The stereocilia on each OHC form a V- or W-shaped pattern, and they arearranged in rows (usually about three) that are graded in height, the tallest stereo-cilia lying on the outside of the V or W. The stereocilia on each IHC are alsoarranged in rows graded in height, but the arrangement is more like a straight lineor a broad arc. In humans, there are about 12 000 OHCs (per ear), each with about140 stereocilia protruding from it, while there are about 3500 IHCs, each with about40 stereocilia.

The tectorial membrane, which has a gelatinous structure, lies above the hair cells.It appears that the stereocilia of the OHCs actually make contact with the tectorialmembrane, but this may not be true for the IHCs. The tectorial membrane appearsto be effectively hinged at one side (the left in Figure 1.6). When the BM movesup and down, a shearing motion is created; the tectorial membrane moves sideways(in the leftright direction in Figure 1.6) relative to the tops of the hair cells. As aresult, the stereocilia at the tops of the hair cells are moved sideways. The movementoccurs via direct contact in the case of the OHCs, but in the case of the IHCs itmay be produced by the viscous drag of fluid streaming between the upper partof the organ of Corti and the tectorial membrane. The movement of the stereociliaof the IHCs leads to a flow of electrical current through the IHCs, which in turnleads to the generation of action potentials (nerve spikes) in the neurones of theauditory nerve. Thus, the IHCs act to transduce mechanical movements into neuralactivity.

The IHCs and OHCs have very different functions. The great majority of afferentneurones, which carry information from the cochlea to higher levels of the auditorysystem, connect to IHCs; each IHC is contacted by about 20 neurones (Spoendlin,1970). Thus, most information about sounds is conveyed via the IHCs. The main roleof the OHCs is actively to influence the mechanics of the cochlea. The OHCs havea motor function, changing their length, shape and stiffness in response to electricalstimulation (Ashmore, 1987; Yates, 1995), and they can therefore influence the re-sponse of the BM to sound. The OHCs are often described as being a key elementin an active mechanism within the cochlea. The function of this active mechanism isdescribed in more detail below.

The action of the OHCs is partly under the control of higher centres of the auditorysystem. There are about 1800 efferent nerve fibres that carry information fromthe auditory system to the cochlea, most of them originating in the superiorolivary complex of the brain stem. Many of these efferent fibres make contact withthe OHCs, and can affect their activity (Liberman and Guinan, 1998). Thus, even theearliest stages in the analysis of auditory signals are partly under the control of highercentres.


IV.2 TUNING ON THE BASILAR MEMBRANE

When the oval window is set in motion by a sound, a pressure difference occursbetween the upper and lower surface of the BM. The pressure wave travels almostinstantaneously through the incompressible fluids of the cochlea. Consequently, thepressure difference is applied essentially simultaneously along the whole length ofthe BM. This causes a pattern of motion to develop on the BM. The pattern does notdepend on which end of the cochlea is stimulated. Sounds which reach the cochlea viathe bones of the head rather than through the air do not produce atypical responses.

The response of the BM to stimulation with a sinusoid takes the form of a travellingwave which moves along the BM from the base towards the apex. The amplitude ofthe wave increases at first and then decreases rather abruptly. The basic form ofthe wave is illustrated in Figure 1.7, which shows schematically the instantaneousdisplacement of the BM for four successive instants in time, in response to a low-frequency sinusoid. The four successive peaks in the wave are labelled 1, 2, 3 and4. This figure also shows the line joining the amplitude peaks, which is called theenvelope. The envelope shows a peak at a particular position on the BM.

The response of the BM to sounds of different frequencies is strongly affectedby its mechanical properties, which vary progressively from base to apex. At thebase the BM is relatively narrow and stiff. This causes the base to respond best tohigh frequencies. At the apex the BM is wider and much less stiff, which causes theapex to respond best to low frequencies. Each point on the BM is tuned; it respondsbest (with greatest displacement) to a certain frequency, called the characteristicfrequency (CF), or best frequency, and responds progressively less as the frequency

Figure 1.7. The solid lines show the instantaneous displacement of the BM at four successiveinstants in time (labelled 14), derived from a cochlear model. The pattern moves from left toright, building up gradually with distance, and decaying rapidly beyond the point of maximaldisplacement. The dashed line represents the envelope traced out by the amplitude peaks in thewaveform.


is moved away from the CF. It is now believed that the tuning of the BM arises fromtwo mechanisms. One is referred to as the passive mechanism. This depends on themechanical properties of the BM and surrounding structures, and it operates in aroughly linear way. The other is the active mechanism. This depends on the operationof the OHCs, and it operates in a nonlinear way. The active mechanism depends onthe cochlea being in good physiological condition, and it is easily damaged. Thetravelling wave shown in Figure 1.7 is typical of what is observed when only thepassive mechanism is operating.

Figure 1.8 shows the envelopes of the patterns of vibration for several different low-frequency sinusoids (data from von Bekesy, 1960). Sounds of different frequenciesproduce maximum displacement at different places along the BM, that is, there is afrequency-to-place transformation. If two or more sinusoids with different frequenciesare presented simultaneously, each produces maximum displacement at its appropriateplace on the BM. In effect, the cochlea behaves like a frequency analyser, althoughwith less than perfect resolution. The resolution is often described in terms of thesharpness of tuning. This refers to the narrowness of the response patterns on the BM.In the case of responses to a single tone, as shown in Figure 1.7, it refers to the spreadof the response along the BM; sharp tuning would be associated with a narrow spread.

Most of the pioneering work on patterns of vibration along the BM was done byvon Bekesy (1960). The vibration patterns found by von Bekesy were rather broad;for example, the pattern for a 400-Hz sinusoid extended along almost the whole lengthof the BM (see Figure 1.8). However, these patterns probably reflect only the passivemechanism. The active mechanism would not have been functioning in von Bekesysexperiments, for two reasons. Firstly, he had to use very high sound levels about140 dB SPL; such high levels are known to damage the active mechanism. Secondly,he used cadaver ears, and the active mechanism ceases to function after death.

Recent work measuring BM responses to sound differs from that of von Bekesy inseveral ways. Firstly, living animals have been used. Great care has been taken to keepthe animals in good physiological condition during the measurements and to minimizetrauma caused by the necessary surgery. Secondly, the techniques themselves aredesigned to be minimally invasive. Finally, rather than measuring the response ofseveral different points on the BM to a single frequency, measurements have usuallybeen made of the responses of a single point to sinusoids of differing frequency. Inthis case, the sharpness of tuning is often measured by adjusting the level at eachfrequency to produce a fixed response on the BM. If a point on the BM is sharplytuned, then the sound level has to be increased rapidly as the frequency is movedaway from the CF. If the tuning is broad, then the sound level has to be increasedonly gradually as the frequency is moved away from the CF. The results show that thesharpness of tuning of the BM depends critically on the physiological condition of theanimal; the better the condition, the sharper is the tuning (Khanna and Leonard, 1982;Sellick, Patuzzi and Johnstone, 1982; Leonard and Khanna, 1984; Robles, Ruggeroand Rich, 1986; Ruggero, 1992; Robles and Ruggero, 2001).

The health of the cochlea is often monitored by placing an electrode in or near theauditory nerve, and measuring the combined responses of the neurones to tone bursts


Figure 1.8. Envelopes of patterns of vibration on the BM for a number of low-frequencysounds. Solid lines indicate the results of actual measurements, while the dashed lines are vonBekesys extrapolations. Redrawn from von Bekesy (1960).

or clicks; this response is known as the compound action potential (AP), or CAP. Thelowest sound level at which an AP can be detected is called the AP threshold. Usually,the BM is sharply tuned when the AP threshold is low, indicating that the cochlea isin good physiological condition and the active mechanism is functioning.


Sou

nd le

vel n

eede

d fo

r cr

iterio

n ve

loci

ty, d

B S

PL

90

80

70

60

50

40

30

20

10

0

1 2 5 10 20Frequency, kHz

Post mortem

Start of experiment

Figure 1.9. Tuning curves measured at a single point on the BM. Each curve shows the inputsound level required to produce a constant velocity on the BM, plotted as a function of stimulusfrequency. The curve marked by open circles was obtained at the start of the experiment when theanimal was in good physiological condition. The curve marked by filled squares was obtainedafter the death of the animal. Data from Sellick, Patuzzi and Johnstone (1982).

An example is given in Figure 1.9, which shows the input sound level (in dBSPL) required to produce a constant velocity of motion at a particular point on theBM, as a function of stimulus frequency (Sellick, Patuzzi and Johnstone, 1982). Thisis sometimes called a constant velocity tuning curve. It is not yet clear whether theeffective stimulus to the IHCs is BM vibration amplitude (equivalent to displacement)or BM velocity. However, for a given frequency, velocity is directly proportional todisplacement: the greater the amplitude, the faster the movement. At the start of theexperiment, when AP thresholds were low, a very sharp tuning curve was obtained(open circles). This curve reflects the contribution of both the passive and the activemechanisms. As the condition of the animal deteriorated, the active mechanism ceasedto function. The tuning became broader, and the sound level required to produce thecriterion response increased markedly around the tip. The broad tuning curve recordedafter death (filled squares) reflects the tuning produced by the passive mechanismalone.

The frequency at the tip of the tuning curve (the CF where the sound level waslowest) shifted downwards when the condition of the animal deteriorated. This oc-curred because, at least for high CFs, the active mechanism gives maximum gain for a


frequency that is somewhat above the best frequency determined by the passive mech-anism. In a healthy ear, the CF is determined mainly by the active mechanism. Whenthe active mechanism ceases to function, the sharply tuned tip of the tuning curve islost and the CF shifts to a lower frequency determined by the passive mechanism.

Even in a healthy ear, the balance between the active and passive mechanismscan change with sound level. If the input level of a sinusoid is held constant, and itsfrequency is varied, then the response of the BM (velocity or amplitude) at a specificpoint shows a peak for a specific frequency. However, the frequency which gives themaximum response often varies with the input sound level (Ruggero et al., 1997).For high CFs, this frequency decreases with increasing sound level, as the relativecontribution of the active mechanism decreases. This implies that, for a fixed inputfrequency, the place on the BM showing the maximum response shifts towards the basewith increasing sound level. Usually, the CF is specified for a low input sound level.

In summary, in a normal healthy ear each point along the BM is sharply tuned,responding with high sensitivity to a limited range of frequencies and requiring higherand higher sound intensities to produce a response as the frequency is moved outsidethat range. The sharp tuning and high sensitivity reflect the active process mediatedby the OHCs.

IV.3 THE NONLINEARITY OF INPUT-OUTPUT FUNCTIONSON THE BASILAR MEMBRANE

In a normal healthy ear, the response of the BM is nonlinear; when the input magnitudeis increased, the magnitude of the response does not grow directly in proportion tothe magnitude of the input (Rhode, 1971; Rhode and Robles, 1974; Sellick, Patuzziand Johnstone, 1982; Robles, Ruggero and Rich, 1986; Ruggero, 1992; Ruggero etal., 1997; Robles and Ruggero, 2001). This is illustrated in Figure 1.10, which showsinput-output functions of the BM for a place with a CF of 8 kHz (from Robles, Ruggeroand Rich, 1986). A series of curves is shown; each curve represents a particularstimulating frequency, indicated by a number (in kHz) close to the curve. The output(velocity of vibration) is plotted on a logarithmic scale as a function of the inputsound level (in dB SPL also a logarithmic scale). If the responses were linear, thefunctions would be parallel to the dashed line. Two functions are shown for a CFtone (8 kHz), one (at higher levels) obtained about one hour after the other. The slightshift between the two was probably caused by a deterioration in the condition of theanimal. An idealized function for a CF tone, with the ordinate scaled in dB units(i.e. as 20log10(velocity)) is shown in Figure 1.11.

While the function for the CF tone is almost linear for very low input sound levels(below 2030 dB) and approaches linearity at high input sound levels (above 90 dB),the function has a very shallow slope at mid-range levels. This indicates a compressivenonlinearity: a large range of input sound levels is compressed into a smaller rangeof responses on the BM. The form of this function can be explained in the followingway. At low and medium sound levels the active mechanism amplifies the responseon the BM. The amplification may be 50 dB or more (Robles and Ruggero, 2001).


Figure 1.10. Input-output functions for a place on the BM with CF = 8 kHz. The stimulatingfrequency, in kHz, is indicated by a number close to each curve. The dashed line indicatesthe slope that would be obtained if the responses were linear (velocity directly proportional tosound pressure). Redrawn from Robles, Ruggero and Rich (1986).

At very low sound levels, below 2030 dB, the amplification is roughly constant andis at its maximal value. As the sound level increases, the amplification progressivelyreduces. Thus, the response grows more slowly than it would in a linear system.When the sound level is sufficiently high, around 90 dB SPL, the active mechanismis unable to contribute any amplification, and the response becomes linear (althoughsome researchers have reported a shallow growth of response even at very high soundlevels). Hence, at high levels, the passive response becomes dominant.

The nonlinearity mainly occurs when the stimulating frequency is close to theCF of the point on the BM whose response is being measured. For stimuli with

Figure 1.11. Schematic input-output function of the BM for a sinusoid at CF. A decibel scaleis used for both axes. The ordinate is scaled (arbitrarily) so that an input of 100 dB gives anoutput of 100 dB. The dashed line shows the slope that would be obtained if the response werelinear.


frequencies well away from the CF, the responses are more linear. Hence, the curvesfor frequencies of 7 and 9 kHz (close to CF) show shallow slopes, while the curves forfrequencies below 7 kHz and above 9 kHz show steeper (linear) slopes. Effectively,the compression occurs only around the peak of the response pattern on the BM. Asa result, the peak in the distribution of vibration along the BM flattens out at highsound levels, which partly accounts for the broad tuning observed by von Bekesy(1960).

Given that the BM response at moderate sound levels is highly nonlinear, onemight expect that, in response to a single sinusoid, harmonics would be generatedand the waveform on the BM would be significantly distorted. In fact, this does notseem to happen (Cooper and Rhode, 1992; Ruggero et al., 1997). The reason whyis not fully understood. Perhaps the active mechanism involves feedback onto theBM in such a way that potential harmonic distortion is filtered out by the passivemechanism. However, in response to inputs containing more than one sinusoidalcomponent, significant distortion can occur. This is described later on in this chapter.

IV.4 TWO-TONE SUPPRESSION

Experiments using two tones have revealed another aspect of nonlinearity on the BM,namely two-tone suppression. The effect is analogous to an effect that was first dis-covered from measurements of the responses of single neurones in the auditory nerve(see below for details). The response to a tone (called the probe tone) with frequencyclose to the CF of the place on the BM being studied can be reduced by a second tonewith a higher or lower frequency, especially when the second tone is higher in levelthan the probe tone (Rhode, 1977; Patuzzi, Sellick and Johnstone, 1984; Ruggero,Robles and Rich, 1992). The effect is illustrated in Figure 1.12 for a probe tone at8.6 kHz (close to the CF) and a suppressor tone at 10.6 kHz. Ruggero, Robles andRich (1992) provided a detailed comparison of the properties of mechanical two-tonesuppression on the BM and two-tone suppression measured in the auditory nerve.They concluded that all of the properties match qualitatively (and mostly quantita-tively) and that two-tone suppression in the auditory nerve probably originates fromtwo-tone suppression on the BM.

IV.5 COMBINATION TONE GENERATION

Another aspect of BM nonlinearity is the generation of distortion products in responseto two or more sinusoidal inputs. These products are often called combination tones.When two sinusoids are presented simultaneously, and their frequency separation isnot too great, their response patterns overlap on the BM. It appears that, at the point ofoverlap, distortion products are generated that behave like additional sinusoidal tones.For example, if the two primary tones presented to the ear have frequencies f1 and f2(f2 > f1), the distortion products have frequencies such as 2f1 f2 and f2 f1. Thedistortion products appear to propagate along the BM to the locations tuned to theirown frequencies (Robles, Ruggero and Rich, 1991). For example, the combination


Figure 1.12. An example of two-tone suppression for a place on the BM with CF = 8.6 kHz.The dashed curve with open circles shows an input-output function for an 8.6-kHz tone alone,referred to as the probe. The solid curves show input-output functions when a suppressor tonewas added to the probe. The suppressor was presented at each of several overall levels, from60 to 90 dB SPL, as indicated by the key in the figure. The solid curves are truncated at thelevel where the suppressor led to an increase in response rather than a decrease. Redrawn fromRuggero, Robles and Rich (1992).

tone with frequency 2f1 f2 produces a local maximum on the BM at the placetuned to 2f1 f2.

Human listeners can sometimes hear these additional tones. The one with frequency2f1 f2 is especially easy to hear. It is audible even for relatively low levels ofthe primary tones when f2 is about 1.2 times f1 (Smoorenburg, 1972a, 1972b). Forexample, if the two primary tones have frequencies 1000 and 1200 Hz, then a tone isheard with frequency 800 Hz. The combination tone with frequency 2f1 + f2 is muchharder to hear. This may be the case because this tone has a higher frequency thanthe two primary tones. Although 2f1 + f2 is probably generated at the point where theresponse patterns of f1 and f2 overlap, it does not propagate to the location on the BMtuned to 2f1 + f2; this would involve propagation in the wrong direction, from theapex towards the base.

IV.6 RESPONSES OF THE BASILAR MEMBRANETO COMPLEX SOUNDS

Consider the response of the BM to two sinusoids, of different frequencies, presentedsimultaneously. Assume that the sinusoids are equal in level, so that two-tone suppres-sion is small. The kind of pattern of vibration that occurs depends on the frequencyseparation of the two sinusoids. If this is very large, then the two sinusoids producetwo, effectively separate, patterns of vibration on the BM. Each produces a maximumat the place on the BM which would have been excited most had that componentbeen presented alone. Thus, the response of the BM to a low-frequency sinusoid of


moderate intensity is essentially unaffected by a high-frequency sinusoid, and viceversa. In this case, the BM behaves like a frequency analyser, breaking down thecomplex sound into its sinusoidal components. Correspondingly, when we listen totwo sinusoids with widely spaced frequencies, we hear two separate tones, with twodifferent pitches. When the two sinusoids are relatively close in frequency, however,the patterns of vibration on the BM interact, so that some points on the BM respondto both of the sinusoids. At those points, the displacement of the BM as a function oftime is not sinusoidal but is a complex waveform resulting from the interference ofthe two sinusoids. When the two sinusoids are sufficiently close in frequency, there isno longer a separate maximum in the pattern of vibration for each of the componentsinusoids; instead there is a single, broader, maximum. Thus, the BM has failed toseparate (resolve) the individual frequency components. Correspondingly, when twosinusoids are very closely spaced in frequency, we cannot hear two separate tones,each with its own pitch; rather, we hear a single sound corresponding to the mixture.This is described more fully in Chapter 3, Section VIII.

Consider now the more complex case of the pattern of responses on the BM toa periodic complex tone, such as a voiced vowel or an instrument playing a note.Such a tone typically contains many harmonics of a common fundamental frequency.For example, the note A4 would have a fundamental component with a frequencyof 440 Hz, and higher harmonics with frequencies of 880, 1320, 1760 . . . Hz. Theharmonics are equally spaced on a linear frequency scale. However, the mappingof CF to distance along the BM roughly follows a logarithmic scale. For example,sinusoids with frequencies of 400, 800, 1600 and 3200 Hz would produce peaks thatwere roughly equally spaced along the BM (see Figure 1.8 above). When a harmoniccomplex tone is presented to the ear, the lower harmonics each give rise to a separatepeak on the BM, while the higher harmonics give responses that overlap, so thatthere are not distinct peaks corresponding to the individual harmonics. A perceptualconsequence of this is that individual low harmonics can often be heard out asseparate tones, while higher harmonics cannot be individually heard; this is describedmore fully in Chapter 3, Section VII. These factors play a crucial role in the perceptionof complex tones, as is explained in Chapter 6, Section II.

IV.7 OTOACOUSTIC EMISSIONS

Evidence supporting the idea that there are active biological processes influencingcochlear mechanics has come from a remarkable phenomenon first reported by Kemp(1978), although predicted by Gold (1948). If a low-level click is applied to the ear,then it is possible to detect sound being reflected from the ear, using a microphonesealed into the ear canal. The early part of this reflected sound appears to come fromthe middle ear, but some sound can be detected for delays from 5 to 60 ms followingthe instant of click presentation. These delays are far too long to be attributed to themiddle ear, and they almost certainly result from activity in the cochlea itself. Thereflected sounds are known as evoked otoacoustic emissions. They have also beencalled Kemp echoes and cochlear echoes.


Although the input click in Kemps experiment contained energy over a wide rangeof frequencies, only certain frequencies were present in the reflected sound. Kempsuggested that the reflections are generated at points on the BM, or in the IHC/OHCtransduction mechanism, where there is a gradient or discontinuity in the mechanicalor electrical properties. The response is nonlinear, in that the reflected sound does nothave an intensity in direct proportion to the input intensity. In fact, the relative level ofthe reflection is greatest at low sound levels; the emission grows about 3 dB for each10 dB increase in input level. This nonlinear behaviour can be used to distinguish theresponse arising from the cochlea from the linear middle ear response. Sometimesthe amount of energy reflected from the cochlea at a given frequency may exceed thatwhich was present in the input sound (Burns, Keefe and Ling, 1998). Indeed, manyears emit sounds in the absence of any input, and these can be detected in the ear canal(Zurek, 1981). Such sounds are called spontaneous otoacoustic emissions, and theirexistence indicates that there is a source of energy within the cochlea which is capableof generating sounds. Kemp (2002) and others have suggested that the emissions area by-product of the active mechanism.

Cochlear emissions can be very stable in a given individual, both in waveform andfrequency content, but each ear gives its own characteristic response. Responses tendto be strongest between 500 and 2500 Hz, probably because transmission from thecochlea back through the middle ear is most efficient in this range, as described earlier.Cochlear emissions can be measured for brief tone bursts as well as clicks, and it iseven possible to detect a reflected component in response to continuous stimulationwith a pure tone.

When the ear is stimulated with two tones, an emission may be detected at thefrequency of one or more combination tones, particularly 2f1 f2. Such emissions arecalled distortion-product otoacoustic emissions. This confirms that the combinationtone is present as a mechanical disturbance in the cochlea, as a travelling wave on theBM.

Sometimes the transient stimulation used to evoke a cochlear echo induces a sus-tained oscillation at a particular frequency, and the subject may report hearing thisoscillation as a tonal sensation. The phenomenon of hearing sound in the absenceof external stimulation is known as tinnitus. It appears that tinnitus may arise fromabnormal activity at several different points in the auditory system, but in a few casesit corresponds to mechanical activity in the cochlea.

In summary, several types of otoacoustic emissions can be identified, includingevoked emissions, spontaneous emissions and distortion-product emissions. While theexact mechanism by which otoacoustic emissions are generated is not understood,there is agreement that it is connected with the active mechanism in the cochlea.

V NEURAL RESPONSES IN THE NORMAL AUDITORY NERVE

Most studies of activity in the auditory nerve have used electrodes with very finetips, known as microelectrodes. These record the nerve impulses, or spikes, in single


auditory nerve fibres (often called single units). The main findings, summarized below,seem to hold for most mammals.

V.1 SPONTANEOUS FIRING RATES AND THRESHOLDS

Most neurones show a certain baseline firing rate, called the spontaneous rate, in theabsence of any external stimulus. Liberman (1978) presented evidence that auditorynerve fibres could be classified into three groups on the basis of their spontaneousrates. About 61 % of fibres have high spontaneous rates (18250 spikes per second),23 % have medium rates (0.518 spikes per second) and 16 % have low spontaneousrates (less than 0.5 spikes per second). The spontaneous rates are correlated with theposition and size of the synapses of the neurones on the IHCs. High spontaneous ratesare associated with large synapses, primarily located on the side of the IHCs facing theOHCs. Low spontaneous rates are associated with smaller synapses on the oppositeside of the IHCs. The spontaneous rates are also correlated with the thresholds of theneurones. The threshold is the lowest sound level at which a change in response ofthe neurone can be measured. High spontaneous rates tend to be associated with lowthresholds and vice versa. The most sensitive neurones may have thresholds close to0 dB SPL, whereas the least sensitive neurones may have thresholds of 80 dB SPL ormore.

V.2 TUNING CURVES AND ISO-RATE CONTOURS

The tuning of a single nerve fibre is often illustrated by plotting the fibres thresholdas a function of frequency. This curve is known as the tuning curve or frequency-threshold curve (FTC). The stimuli are usually tone bursts, rather than continuoustones. This avoids the effects of long-term adaptation (a decrease in response over timethat can occur with continuous stimulation), and also makes it easier to distinguishspontaneous from evoked neural activity. The frequency at which the threshold of thefibre is lowest is called the characteristic frequency (CF) (the same term is used todescribe the frequency to which a given place on the BM is most sensitive). Sometypical tuning curves are presented in Figure 1.13. On the logarithmic frequencyscale used, the tuning curves are usually steeper on the high-frequency side than onthe low-frequency side. It is generally assumed that the tuning seen in each singleauditory nerve fibre occurs because that fibre responds to activity in a single IHC at aparticular point on the BM. Iso-velocity tuning curves on the BM are similar in shapeto neural FTCs (Khanna and Leonard, 1982; Sellick, Patuzzi and Johnstone, 1982;Robles, Ruggero and Rich, 1986; Ruggero et al., 1997).

The CFs of single neurones are distributed in an orderly manner in the auditorynerve. Fibres with high CFs are found in the periphery of the nerve bundle, and there isan orderly decrease in CF towards the centre of the nerve bundle (Kiang et al., 1965).This kind of arrangement is known as tonotopic organization and it indicates that theplace representation of frequency along the BM is preserved as a place representationin the auditory nerve.


Figure 1.13. A sample of tuning curves (also called frequency-threshold curves) obtainedfrom single neurones in the auditory nerve of anaesthetized cats. Each curve shows results forone neurone. The sound level required for threshold is plotted as a function of the stimulusfrequency (logarithmic scale). Redrawn from Palmer (1987).

In order to provide a description of the characteristics of single fibres at levelsabove threshold, iso-rate contours can be plotted. To determine an iso-rate contour,the intensity of sinusoidal stimulation required to produce a predetermined firing ratein the neurone is plotted as a function of frequency. The resulting curves are generallysimilar in shape to tuning curves, although they sometimes broaden at high soundlevels. Also, for high CFs, the frequency at the tip (the lowest point on the curve)may decrease slightly with increases in the predetermined firing rate. This reflects thechange in BM tuning with level described earlier.

V.3 RATE-VERSUS-LEVEL FUNCTIONS

Figure 1.14 shows schematically how the rate of discharge for three auditory nervefibres changes as a function of stimulus level. The curves are called rate-versus-levelfunctions. In each case, the stimulus was a sinusoid at the CF of the neurone. Considerfirst the curve labelled (a). This curve is typical of what is observed for neurones withhigh spontaneous firing rates. Above a certain sound level the neurone no longerresponds to increases in sound level with an increase in firing rate; the neurone is saidto be saturated. The range of sound levels between threshold and the level at whichsaturation occurs is called the dynamic range. For neurones with high spontaneousrates, this range is often quite small, about 1530 dB. Curve (b) is typical of whatis observed for neurones with medium spontaneous rates. The threshold is slightlyhigher than for (a) and the dynamic range is slightly wider. Curve (c) is typical of whatis observed for neurones with low spontaneous rates. The threshold is higher than for(b). The firing rate at first increases fairly rapidly with the increasing sound level, but


Figure 1.14. Schematic examples of how the discharge rates of single auditory neurones varyas a function of stimulus level. The curves are called rate-versus-level functions. In each case,the stimulus was a sinusoid at the CF of the neurone. Curves (a), (b) and (c) are typical of whatis observed for neurones with high, medium and low spontaneous firing rates, respectively.

then the rate of increase slows down. The firing rate continues to increase graduallywith increasing sound level over a wide range of levels. This has been called slopingsaturation (Sachs and Abbas, 1974; Winter, Robertson and Yates, 1990).

The shapes of rate-versus-level functions can be understood in terms of two func-tions (Yates, 1990; Patuzzi, 1992). This is illustrated in Figure 1.15. The first functionis the input-output function of the BM, illustrated schematically in the top-right panel.The second is the function relating the spike rate in a specific neurone to the magnitudeof the BM response. This second function is similar in form for different neurones,showing saturation when the BM amplitude is a certain factor above the value requiredfor threshold, but it varies in the magnitude required for threshold. Three such func-tions are illustrated schematically in the top-left panel of Figure 1.15. The rate-versus-level functions corresponding to these three functions are shown in the bottom-rightpanel.

The variation across neurones depends mainly on the type of synapse, as dis-cussed earlier. Neurones with low thresholds have large sensitive synapses. Theystart to respond at very low sound levels, where the input-output function on theBM is nearly linear. As the sound level increases, the BM displacement increases


Figure 1.15. Schematic illustration of how the shapes of rate-versus-level functions can beaccounted for in terms of the BM input-output function (top-right panel), and the functionsrelating neural firing rate (APs/sec) to amplitude on vibration on the BM (top-left panel). Threesuch functions are shown, corresponding to synapses with different sensitivities. The resultingthree rate-versus-level functions are shown in the bottom panel. Adapted from Patuzzi (1992).

in a nearly linear manner, and the neurone saturates relatively early, giving asmall dynamic range, as shown by the left-most curve in the lower panel. Neu-rones with higher thresholds have less sensitive synapses. They respond over therange of sound levels where the BM input-output function shows a strong com-pressive nonlinearity. Hence, a large increase in sound level is needed to increasethe BM displacement to the point where the neurone saturates, and the neu-rone has a wide dynamic range, as shown by the right-most curve in the lowerpanel.

V.4 TWO-TONE SUPPRESSION

Auditory neurones show an effect that is exactly analogous to the two-tone suppressionon the BM that was described earlier. Indeed, the neural effect was discovered longbefore the BM effect. The tone-driven activity of a single fibre in response to one tonecan be suppressed by the presence of a second tone. This was originally called two-tone inhibition (Sachs and Kiang, 1968), although the term two-tone suppression isnow generally preferred, since the effect does not appear to involve neural inhibition.Typically the phenomenon is investigated by presenting a tone at, or close to, the CF ofa neurone. A second tone is then presented, its frequency and intensity are varied andthe effects of this on the response of the neurone are noted. When the frequency and


80

60

40

20

1.5 5 10 15Frequency, kHz

Figure 1.16. Neurophysiological data from Arthur, Pfeiffer and Suga (1971). The open circlesshow the tuning curve (threshold versus frequency) of a single neurone with a CF at 8 kHz.The neurone was stimulated with a tone at CF, and just above threshold (indicated by the opentriangle). A second tone was then added and its frequency and intensity were varied. Any tonewithin the shaded areas bounded by the solid circles reduced the response to the tone at CF by20 % or more. These are the suppression areas.

intensity of the second tone fall within the excitatory area bounded by the tuning curve,this usually produces an increase in firing rate. However, when they fall just outsidethat area, the response to the first tone is reduced or suppressed. The suppression isgreatest when the suppressor falls in one of two frequency regions on either side of theexcitatory response area, as illustrated in Figure 1.16. The suppression begins veryquickly when the suppressor is turned on, and ceases very quickly when it is turnedoff (Arthur, Pfeiffer and Suga, 1971). This is consistent with the likely origin of thesuppression as a mechanical effect on the BM.

V.5 PHASE LOCKING

In response to a sinusoid with a frequency below about 5 kHz, the nerve firingstend to be phase locked, or synchronized, to the evoked waveform on the BM. Agiven nerve fibre does not necessarily fire on every cycle of the stimulus but, whenfirings do occur, they occur at roughly the same phase of the waveform each time.Thus, the time intervals between firings are (approximately) integer multiples ofthe period of the stimulating waveform. For example, a 500-Hz tone has a periodof 2 ms; the waveform repeats regularly every 2 ms. The intervals between nervefirings in response to a 500-Hz tone are approximately 2, or 4, or 6, or 8 ms, and soon. Neurones do not fire in a completely regular manner, so that there are not exactly500, or 250 or 125 spikes/s. However, information about the period of the stimulating


Figure 1.17. Interspike interval histograms for a single auditory neurone (in the squirrel mon-key) with a CF of 1.6 kHz. The frequency of the sinusoidal input and the mean response rate inspikes per second are indicated above each histogram. All tones had a level 80 dB SPL. Noticethat the time scales in E and F differ from those in A to D. Redrawn from Rose et al. (1968).

waveform is carried unambiguously in the temporal pattern of firing of a singleneurone.

One way to demonstrate phase locking in a single auditory nerve fibre is to plot ahistogram of the time intervals between successive nerve firings. Several such inter-spike interval histograms for a neurone with a CF of 1.6 kHz are shown in Figure 1.17.For each of the different stimulating frequencies (from 0.408 to 2.3 kHz in this case),the intervals between nerve spikes lie predominantly at integer multiples of the periodof the stimulating tone. These intervals are indicated by dots below each abscissa.Thus, although the neurone does not fire on every cycle of the stimulus, the distribu-tion of time intervals between nerve firings depends closely on the frequency of thestimulating waveform.

Phase locking does not occur over the whole range of audible frequencies. In mostmammals it becomes progressively less precise for stimulus frequencies above 1 kHz,and it disappears completely at about 45 kHz (Rose et al., 1968), although the exactupper limit varies somewhat across species (Palmer and Russell, 1986). Phase lockingimproves in precision with increasing sound level at low levels, and then stays roughlyconstant in precision over a very wide range of sound levels.

Information from phase locking contributes to the ability to localize sounds inspace, and it probably also plays a role in the perception of pitch and in the perception


of speech when background sounds are present. These aspects of auditory perceptionare discussed more fully in Chapters 5, 7 and 8.

VI TYPES OF HEARING LOSS

A conductive hearing loss is caused by a reduced efficiency of sound transmissionthrough the outer and/or middle ear. This may be caused by wax (cerumen) in the earcanal, damage to the eardrum produced by infection or trauma, damage to or stiffeningof the ossicles in the middle ear, or fluid in the middle ear caused by infection. It resultsin an attenuation of sound reaching the cochlea, so that sounds appear quieter thannormal. The amount of loss may vary with frequency, so sounds may appear to have asomewhat different tonal quality from normal. However, these are the main perceptualconsequences of a conductive loss; unlike cochlear hearing loss, it does not generallyresult in marked distortions or abnormalities in other aspects of sound perception.Conductive hearing loss can often be treated by drugs (to cure infections) or surgery.

Cochlear hearing loss involves damage to the structures inside the cochlea. It canarise in many ways, for example by exposure to intense sounds or to ototoxic chemicals(such as certain antibiotics, drugs used to treat high blood pressure or solvents), byinfection, by metabolic disturbances, by some forms of allergies, by autoimmunedisorders and as a result of genetic factors. These agents can produce a variety oftypes of damage to the cochlea, and, to complicate matters further, the damage mayextend beyond the cochlea. For example, an infection may produce damage at severalsites, such as the auditory nerve and higher centres in the auditory pathway. Whenboth cochlear and neural structures are involved, the more general term sensorineuralhearing loss is used.

Finally, hearing loss can occur through damage to structures or neural systemsoccurring at a level in the auditory system beyond the cochlea, for example in theauditory nerve or the auditory cortex. Such types of hearing loss are given the generalname retrocochlear hearing loss. A relatively common cause of retrocochlear lossis the growth of a benign tumour (often called an acoustic neuroma or vestibularschwannoma) which presses on the auditory nerve.

Although sensorineural hearing loss can involve structures other than the cochlea,it is common for the most serious damage to occur within the cochlea. This is probablytrue for the majority of cases of presbyacusis, the hearing loss that is associated withageing. Furthermore, it is common for the damage to be largely confined to certainspecific structures within the cochlea. This book is concerned with cases where thehearing loss arises primarily from damage to the cochlea. Henceforth, when the phrasehearing impairment is used, it should be taken to imply hearing impairment causedby cochlear damage. However, it should be borne in mind that, while it is relativelyeasy to produce pure cochlear damage in animal models, such pure damage maybe relatively rare in hearing-impaired humans.

In the last twenty years there has been a considerable increase in understandingof the physiology and function of the normal cochlea. Along with this has come an


improved understanding of the changes in function that are associated with cochlearhearing loss. The rest of this chapter reviews the structure and function of the impairedcochlea.

VII PHYSIOLOGY OF THE DAMAGED COCHLEA

There is strong evidence that the functioning of the normal cochlea depends upon theoperation of an active mechanism that is linked to the integrity of the OHCs. Thismechanism may involve feedback of energy onto the BM, via the OHCs, and it playsan important role in producing the high sensitivity of the BM to weak sounds and thesharp tuning on the BM. The normal BM has a strongly nonlinear response, whichresults in compressive input-output functions, two-tone suppression and combination-tone generation.

Cochlear hearing loss often involves damage to the OHCs and IHCs; the stereociliamay be distorted or destroyed, or entire hair cells may die. The OHCs are generallymore vulnerable to damage than the IHCs. Some examples of OHC damage are shownin Figure 1.18. When OHCs are damaged, the active mechanism tends to be reducedin effectiveness or lost altogether. The function of the OHCs can also be adverselyaffected by malfunctioning of the stria vascularis (Schmiedt, 1996). As a result, severalchanges occur: the sensitivity to weak sounds is reduced; so sounds need to be moreintense to produce a given magnitude of response on the BM, the tuning curves on theBM become much more broadly tuned and all of the frequency-selective nonlineareffects weaken or disappear altogether.

VII.1 BASILAR MEMBRANE RESPONSES

There have been many studies showing that the responses of the BM are highlyphysiologically vulnerable. One example has already been given; see Figure 1.9.Generally, the tuning on the BM becomes less sharp, and the sensitivity around thetip is reduced, when the cochlea is damaged. In the great majority of studies, thechanges in BM responses have been associated with some form of damage to theOHCs, either directly or via some form of metabolic disturbance.

The effects of cochlear damage on the input-output functions of the BM of a chin-chilla are illustrated in Figure 1.19 (Ruggero and Rich, 1991). The solid curve withblack squares, labelled Before, shows the input-output function obtained when thecochlea was in good condition; the stimulus was a sinusoid with frequency corre-sponding to the CF of 9000 Hz. The curve shows a compressive nonlinearity for inputsound levels between about 30 and 90 dB SPL. In contrast, the response to a sinusoidwith a frequency of 1000 Hz, well below the CF, is steeper and is almost linear (solidcurve with open circles).

To manipulate the functioning of the cochlea, the animal was injected withfurosemide (also known as frusemide), a diuretic that is known to disrupt hair cellpotentials, and hence disrupt the function of the hair cells. The dashed curves in


Figure 1.18. Examples of patterns of damage to the OHCs. For the top two panels, the damagewas caused by the administration of aminoglycosides. For the bottom panel, the damage wasof unknown origin. In the upper panel, the stereocilia of some OHCs are missing. In the middlepanel, the stereocilia of the OHCs are mostly either missing or grossly deformed. In the bottompanel, the stereocilia of the OHCs are completely missing over a certain region. The bottompanel also shows expansion of the supporting cells to replace lost hair cells. The electronmicrographs were supplied by Dr Andrew Forge of the Institute of Laryngology and Otology,University College London Medical School.


Figure 1.19. Input-output functions on the BM immediately preceding (solid lines) and fol-lowing (broken lines) an intravenous furosemide injection. See text for details. Redrawn fromRuggero and Rich (1991).

Figure 1.19 were obtained at various times after injection of the drug; the time isindicated by a range in minutes next to each curve. Shortly after the injection (1119minutes), the input-output function for the CF tone was markedly altered. The biggestalteration was at low sound levels. To produce a given response on the BM (say, 40m/s), the input level had to be in

Documents

Cochlear Hearing Loss: Physiological, Psychological and Technical Issues